Corollary 9·1 — (w_t − κ₁n_t)/(σ(t/μ₁)^{1/2}) and, if μ₂ < ∞, (w_t − (κ₁/μ₁)t)/(γt/μ₁)^{1/2} are asymptotically standard normal
ProvedSmithRegenerative.CLT.corollary_9_1central-limit-theoremcumulative-processp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1regenerative-processrenewal-reward
Let be a cumulative process on a renewal process with and , with cycle increments , variation increments , and the number of regenerations in . Suppose and , and put
Then:
- for every real ,
- if in addition , and is positive, then for every real ,
The second assertion is the central limit theorem for renewal-reward processes with a deterministic centring; applied to it gives Feller's central limit theorem for renewal processes.
Formalization Note and are implicit in the paper, which divides by and . The two assertions are the two conjuncts of the statement.
Preamble
import Mathlib import Definitions.Def_SmithRegenerative_CLT_CumulativeProcess open MeasureTheory ProbabilityTheory Filter Topology
Formal statement
namespace SmithRegenerative.CLT
/-- Smith (1955), Corollary 9·1, p. 30: "If `w_t` is a general cumulative process for which
`E(Δ_n w̃_t)² < ∞`, `κ₁ ≡ EΔ_n w_t`, `κ₂ ≡ E(Δ_n w_t)² < ∞`, and `σ² ≡ κ₂ − κ₁²`, then provided
`μ₁ < ∞`, `lim_{t=∞} P{(w_t − κ₁n_t)/(σ(t/μ₁)^{1/2}) ≤ α} = Φ(α)` (5·4·8). If, in addition,
`μ₂ < ∞`, then `lim_{t=∞} P{(w_t − (κ₁/μ₁)t)/((γt/μ₁)^{1/2}) ≤ α} = Φ(α)` (5·4·9), where
`γ = var Δ_n(w_t − κ₁μ₁⁻¹t)`."
For a cumulative process with `t₀ = 0`, `w₀ = 0` (the model with `M = 1`); the two assertions
are the two conjuncts, the second under `μ₂ = E t₁² < ∞`.
**Formalization Note** `σ > 0` and `γ > 0` are implicit (the paper divides by `σ` and `√γ`).
`Δ_n(w_t − κ₁μ₁⁻¹t) = y_n − (κ₁/μ₁) t_n`, so `γ = var(y₁ − (κ₁/μ₁) t₁)`. `n_t` is `count`.
`Φ = cdf (gaussianReal 0 1)`, and each limit is asserted for every real `α`. -/
theorem corollary_9_1 {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
(τ : ℕ → Ω → ℝ) (w : ℝ → Ω → ℝ) (hw : IsCumulativeModel P τ (fun _ : Fin 1 => w))
(hμ : Integrable (τ 1) P)
(hvar2 : Integrable (fun ω => varIncr w τ 1 ω ^ 2) P)
(σ : ℝ) (hσ_pos : 0 < σ)
(hσ : σ ^ 2 = (∫ ω, incr w τ 1 ω ^ 2 ∂P) - (∫ ω, incr w τ 1 ω ∂P) ^ 2) :
(∀ α : ℝ, Tendsto
(fun t : ℝ => P.real {ω | (w t ω - (∫ ω', incr w τ 1 ω' ∂P) * (count τ t ω : ℝ)) /
(σ * Real.sqrt (t / mu1 P τ)) ≤ α})
atTop (𝓝 (cdf (gaussianReal 0 1) α))) ∧
(Integrable (fun ω => τ 1 ω ^ 2) P →
∀ γ : ℝ, 0 < γ →
γ = variance (fun ω => incr w τ 1 ω - (∫ ω', incr w τ 1 ω' ∂P) / mu1 P τ * τ 1 ω) P →
∀ α : ℝ, Tendsto
(fun t : ℝ => P.real {ω | (w t ω - (∫ ω', incr w τ 1 ω' ∂P) / mu1 P τ * t) /
Real.sqrt (γ * t / mu1 P τ) ≤ α})
atTop (𝓝 (cdf (gaussianReal 0 1) α))) := by sorry
end SmithRegenerative.CLT
Source
Smith, Regenerative stochastic processes, Proc. R. Soc. Lond. A 232(1188):6–31 (1955), DOI 10.1098/rspa.1955.0198, p. 30, Corollary 9·1 (5·4·8)–(5·4·9)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.